What problem does it solve? When a delivery agent needs to know something about the world outside its own system — prior art, what a standard says, how practitioners solved a similar problem — it needs a disciplined way to search, weigh, and cite external sources rather than guessing from training data. ## Core Features & Use Cases - Five-perspective coverage: Works through primary sources, practitioner consensus, recent developments, counter-perspectives, and cross-domain analogies for every question. - Source-tier weighting: Ranks evidence from peer-reviewed and official sources above community content, and calibrates stated confidence to the tier actually fetched. - Honest degraded mode: If no external sources are reachable, it declares that in the first line and labels findings as model knowledge instead of returning an empty report. - Use Case: Dispatched by sdlc-explore to answer "how do other teams version event schemas?", it searches, fetches, cites real URLs, deduplicates its own findings, and returns a conclusion-first JSON answer the caller synthesises into explore.md. ## Quick Start Dispatch this skill with one external research question and receive a cited, conclusion-first answer as a JSON text block.
